What if your Google Shopping impressions fell by about a quarter while your CTR rose by nearly 30%, and that wasn’t a bug but the new normal?
We have spent the last few weeks inside client accounts watching something strange happen. Fewer people are clicking our Shopping ads. The ones who do click are converting more often. And our first instinct (panic, then a bid reset) would have been exactly the wrong move.
Here is what is actually going on.
What is the reverse crocodile effect?
Why would impressions fall while CTR improves?
The answer appears to be a narrower, more selective paid search funnel.
In organic search, the familiar “crocodile effect” describes impressions rising while clicks decline. AI Overviews and AI Mode can expose a brand to more searches without generating a proportional increase in visits.
Shopping is showing the opposite pattern.
Industry analysis from Search Engine Roundtable and Search Engine Land highlights a “reverse crocodile” pattern:
- Median Shopping impressions fell from approximately 1.85 million to 1.4 million per account.
- Median CTR increased from roughly 1.20% to nearly 1.55%.
- The analysis covered approximately 175 billion Shopping ad impressions.
- Click volume remained broadly flat or declined slightly.
That is approximately a 24% reduction in median impressions alongside a 29% increase in CTR.
The crucial detail? A higher CTR does not necessarily mean more demand. It may simply mean Google is showing your product to fewer, more commercially relevant shoppers.
The old funnel was wide at the top: broad searches, comparison queries and low-intent product discovery. AI Mode can now absorb some of that research journey inside the search experience. The remaining Shopping placements are more likely to appear when the shopper is closer to making a decision.
What should we measure instead?
Do not evaluate this shift through CTR alone. Track:
- Clicks
- Conversion rate
- Revenue
- Contribution margin
- Impression share
- New-customer acquisition cost
- Assisted conversion value
If CTR rises but clicks and revenue fall, the account may be losing reach.
If impressions fall, CTR rises and conversion rate improves, the account may be becoming more efficient, but only if revenue and margin support the conclusion.

Why does AI Mode change the acquisition math?
What happens when Google answers the early part of the buying journey before a shopper clicks an ad?
AI Mode is designed to interpret conversational queries, compare products and surface recommendations within a generated answer. Google has also announced Shopping experiences that can place product carousels, sponsored products and offers inside AI-led search journeys. Its AI-powered Shopping announcement provides useful context on how these experiences are developing.
This changes the role of the Shopping feed.
Previously, a product could compete through a combination of bid, title, image, price and historical performance. Those signals still matter. But AI-led discovery adds another question:
Can the system understand why this product is suitable for this specific shopper?
A thin product feed might say:
“Premium waterproof jacket.”
A richer feed explains:
- Waterproof rating
- Breathability
- Material
- Temperature range
- Intended activity
- Fit
- Available sizes
- Care instructions
- Review count
- Compatible accessories
The second product gives an AI system more evidence to work with. It can match the item to a natural-language request such as “a lightweight waterproof jacket for wet autumn hikes under £150.”
That is why product feeds have become a form of ad creative. The feed does not just help Google identify what you sell. It helps Google decide when your product belongs in the answer.
What should we do this week?
Audit your highest-value products in Merchant Center. For each SKU, check whether the feed communicates:
- What the product is.
- Who it is for.
- Where and when it should be used.
- How it compares with alternatives.
- Why it is worth the price.
Bear in mind, richer data is not a licence to add unsupported claims. Every attribute must match the landing page, packaging and product reality.
Why are some verticals feeling more pressure than others?
Why might home goods and electronics accessories struggle while apparel remains more stable?
Our August 2026 account review showed an uneven acquisition shift:
- Home goods: customer acquisition cost rose noticeably
- Electronics accessories: CAC rose more sharply still
- Apparel and footwear with strong UGC and detailed fit attributes: CAC stable or declining
These are observations from our client cohort, not a universal industry benchmark. The difference appears to be linked to how easily an AI system can evaluate product suitability.
Home goods often involve nuanced dimensions, materials, room fit and installation requirements. Electronics accessories depend on compatibility, model numbers, charging standards and use cases. If that information is missing or inconsistent, the product becomes harder to recommend confidently.
Apparel has its own complexity, but brands with strong user-generated content, accurate size and fit data, detailed materials and reliable reviews can give both shoppers and AI systems stronger decision signals.
What is the technical takeaway?
Create a feed completeness score for priority SKUs.
For example:
Feed completeness =
completed required attributes
+ completed decision attributes
+ verified review data
+ landing-page consistency
You can weight the score by margin or revenue potential. A best-selling product with an 82% completeness score deserves attention before a low-margin SKU with a perfect feed.
For electronics, prioritise compatibility and model identifiers.
For home goods, prioritise dimensions, materials, installation and room use cases.
For apparel, prioritise fit, material, size availability, reviews and high-quality imagery.

Is last-click ROAS now misleading?
What if AI Mode influences the sale but never receives the final click?
This is the attribution trap.
A shopper may:
- Discover a product through an AI-led Shopping experience.
- Compare it with two alternatives.
- Leave without clicking.
- Return through a branded search three days later.
- Convert through the brand website.
Last-click reporting credits branded search. The earlier Shopping exposure disappears from the commercial story.
We see this regularly during signal quality audits. In one recent account, we found conversions that the off-the-shelf setup was not reporting correctly. The fix was not more budget. It was better measurement. You can read the technical breakdown in our signal quality audit.
Google does not currently provide a clean, universal reporting segment that separates AI Mode delivery from conventional Shopping or Performance Max delivery. AI-influenced activity is therefore best evaluated through a combination of:
- Data-driven attribution
- Assisted conversion paths
- Conversion lag
- Branded search lift
- New-customer revenue
- Incrementality testing
- Blended ROAS
How can we build a practical proxy?
Keep auto-tagging enabled so Google Click IDs pass into your analytics system. Then use consistent UTMs for campaigns designed to capture AI-led and broader Shopping demand:
utm_source=google
utm_medium=cpc
utm_campaign=shopping_ai_ready
utm_content=priority_feed
This will not isolate every AI Mode impression. It will, however, make campaign-level comparisons cleaner in GA4 and help you understand whether feed improvements are increasing qualified traffic and revenue.
Our attribution evolution analysis covers why single-touch reporting is becoming less useful as buying journeys become more fragmented.

What are the four moves for this week?
How should an ecommerce team respond without overreacting?
1. Audit Merchant Center completeness
Start with your top 20% of products by revenue or margin. Review title, description, material, dimensions, use case, compatibility, reviews, shipping, returns, price and availability.
Fix inconsistencies before adding budget.
2. Watch impression share, not impressions alone
A raw impression decline may reflect weaker demand, tighter eligibility or AI-led pruning. Compare it with impression share, lost impression share due to budget and lost impression share due to rank.
If impressions fall but impression share remains healthy, the market may simply be producing fewer eligible searches.
3. Move from channel ROAS to blended ROAS
Include direct revenue, branded search lift and assisted conversions where possible. Compare performance over a consistent 30- or 60-day window to account for conversion lag.
A campaign that appears weak on last-click ROAS may be supporting a profitable branded search engine.
4. Treat each impression as a premium asset
When fewer Shopping ads appear, every eligible placement carries more weight. Product imagery, price clarity, promotions, reviews and landing-page speed all matter more.
Our commission-based growth case study shows why commercial alignment matters when performance is judged against real revenue rather than surface-level metrics.
The takeaway: being chosen matters more than being present
The reverse crocodile is not necessarily a glitch. It is a signal that Google Shopping is moving away from cheap, broad visibility and toward AI-mediated product selection.
Fewer impressions can be healthy if they represent better-qualified demand. Higher CTR can be misleading if it is simply the result of a shrinking denominator. And last-click ROAS can understate the value of the campaigns introducing your products to shoppers before they are ready to buy.
The strategic priority is clear:
- Make your product data easier for AI systems to understand.
- Measure revenue across the full journey.
- Protect impression share on high-value products.
- Judge performance through profit, not CTR in isolation.
How are your Shopping impressions trending this quarter: up, down or misleading? And what does your blended ROAS say compared with your last-click number?
Category: Positive Sparks News